Distracted Driver Detection Using Convolutional Neural Networks Based Segmentation Model

In recent years, there has been a significant increase in road accidents caused by driver distraction. Therefore, researchers have proposed various approaches to solving this problem. Machine learning and deep learning models have recently been the most widely used solutions for detecting driver behavior. However, most methods use whole image as input data, while only certain regions contain relevant information. To solve this problem and improve the performance of distracted driver detection, we introduce a two-stage model; the first stage identifies the region of interest that contains the driver using image segmentation, while the second stage focuses on feature extraction and classification of driver distraction. Our model achieves competitive results, with an accuracy of 99.37% on the New AUC dataset and 99.82% on the State Farm dataset.